Introduction

Basketball is a high-intensity, intermittent team sport marked by frequent transitions between offense and defense, requiring players to perform repeated high-intensity actions such as sprints, jumps, and changes of direction (COD) during games1,2. The capacity to repeatedly perform these explosive movements with minimal fatigue is essential for sustaining performance throughout a match. Repeated sprint ability (RSA) is recognized as a key fitness component in team sports and is defined as the capacity to perform multiple short-duration sprints (≤ 10 s) with brief recovery intervals (≤ 60 s)3. In basketball, players frequently perform short sprints during games, most of which involve COD4. The significance of RSA in basketball is underscored by its association with match performance and its capacity to distinguish players across competitive levels5. Recent studies have focused on designing effective training methods to improve RSA in team sports athletes.

Repeated sprint training (RST) has been developed to improve RSA, traditionally relying on straight-line sprints6. However, considering the multidirectional nature of basketball, incorporating COD elements into RST may better align with the sport’s demands and increase training specificity7. Research shows that RST with COD enhances both linear and multidirectional sprint performance8, an adaptation especially relevant to basketball, where most sprints require directional changes5. Moreover, compared with linear sprints, COD movements place greater metabolic demands, trigger distinct muscle activation patterns, and impose unique stresses on the neuromuscular system during rapid deceleration and acceleration phases4,9,10. Under these loading conditions, COD training can induce specific neuromuscular adaptations, including enhanced motor unit recruitment, improved intermuscular coordination, increased eccentric strength, and more efficient use of the stretch–shortening cycle. Recent evidence further demonstrates that combining COD with plyometric training produces significant neuromuscular adaptations in female basketball players, including improvements in sprint performance, agility, jump height, and RSA11,12. These physiological and mechanical differences highlight the potential value of COD-specific training.

Given these unique demands and potential benefits, researchers have examined how incorporating COD into RST affects performance outcomes in team sport athletes13. However, findings are inconsistent: some studies report greater adaptations from COD training than from linear sprinting8, whereas others show comparable improvements14. These inconsistencies may stem from variations in training design, participant characteristics, and assessment methods. Moreover, the nature of the training tasks themselves can shape adaptation outcomes. For instance, complex possession-based drills have been shown to improve both physical and cognitive performance in young athletes, suggesting that training design and task complexity play a decisive role in shaping the type and magnitude of adaptations. This highlights the need for further research15. RSA development is influenced by both metabolic factors (rapidly resynthesize phosphocreatine, buffer hydrogen ions (H+), and maintain oxidative capacity) and neural factors (activate and recruit muscle fibers)3. RST with COD may provide additional benefits by enhancing neuromuscular coordination and facilitating transition between movement patterns10. Despite growing interest, research on female athletes is still limited. A recent study of female basketball players reported that RST with COD improved multiple physical performance measures, although optimal training parameters remain unclear16. This gap is critical because female athletes may respond differently to high-intensity training due to physiological differences, such as lower muscle mass and distinct hormonal profiles. These considerations justify the need for gender-specific investigations.

Given the importance of RSA in basketball and the limited research on female athletes, particularly regarding COD-specific training, there is a need to investigate the effectiveness of different RST protocols in this population. Therefore, the purpose of this study was to examine the effects of a short-term RST program with and without COD on physical performance in collegiate female basketball players. We hypothesized that COD-based RST would lead to greater improvements in repeated sprint ability (RSAave and RSAFI), maximal oxygen uptake (VO2max), and change-of-direction speed (zig–zag test), whereas no significant between-group differences would be observed in short-distance sprint times (10–30 m), single-effort RSA (RSAbest), or countermovement jump (CMJ) height.

Methods

Experimental approach to the problem

A randomized parallel-group design was employed to investigate the effects of a 6-week intervention involving RST with COD and linear RST (administered two times per week) on sprint, neuromuscular, RSA, aerobic capacity, and COD in female collegiate basketball players. To ensure unbiased participant allocation, randomization was performed using a computer-generated sequence. Participants were then assigned to either the COD (EXP) or linear (CON) repeated sprint group. The sprint intervention programs were incorporated into the standard basketball training regimen of each group during the pre-season period. Two familiarization sessions were conducted before baseline testing to ensure participants understood the protocol. The tests battery included: (a) 10 m sprint, (b) 20 m sprint, (c) 30 m sprint, (d) vertical jump, (e) RSA, (f) maximal oxygen uptake, and (g) zig–zag test. Participants were instructed to maintain their usual lifestyle and diet. Testing occurred after 72 h of recovery from intense physical activity. Baseline and post-testing sessions were conducted between 6 and 8 PM under similar environmental conditions to minimize circadian rhythm effects. Participants wore standardized athletic attire during all testing sessions.

Participants

The sample size was calculated using Gpower software (v3.1 Düsseldorf University, Germany) with an Cohen’s f effect size of 0.4, significance level α of 0.05, and statistical power of 0.817, indicating a minimum requirement of 15 participants per group. A total of 30 trained female college basketball players (age: 22.40 ± 4.40 years, weight: 65.10 ± 5.60 kg, height: 179.60 ± 5.00 cm) voluntarily participated in this study. All participants met the following inclusion criteria: age ≥ 18 years, at least 5 years of structured basketball training experience, including youth-level school or club training and collegiate or semi competitive training, while excluding purely recreational or non-structured participation, and training frequency ≥ 3 times per week with sessions lasting ≥ 2 h Notably, participants were classified as Tier 2 training level, which indicates athletes with several years of systematic training experience and stable training frequency, who typically compete at the collegiate or developmental level but have not yet reached national or international elite status18. Exclusion criteria included recent lower limb injury or surgery, engagement in other sports, cardiovascular or pulmonary diseases, or use of performance-affecting medications. Participants provided informed consent, and the study was approved by the ethics committee of university adhering to national and international human experiment regulations.

Testing procedures

To minimize potential learning effects, participants completed two structured familiarization sessions one week prior to baseline testing. During these sessions, participants practiced all physical performance tests under research staff supervision to ensure correct execution and consistency. This procedure aimed to standardize movement patterns and reduce variability associated with inexperience in testing protocols. All physical tests were completed in a single session, scheduled one week before the start of the training program, in a controlled environment that included both a laboratory and a basketball court. A fixed testing sequence was followed to minimize fatigue effects and ensure the accuracy of performance outcomes. The sequence included sprint tests (10 m, 20 m, 30 m), countermovement jump (CMJ), zig–zag COD test, repeated sprint ability (RSA) test, and maximal oxygen uptake (VO2max) test. This sequence was structured to progress from tests emphasizing neuromuscular and short durations demands to those requiring greater metabolic involvement. Appropriate recovery intervals were provided between tests to minimize residual fatigue. Each testing session lasted approximately 90–120 min and was supervised by experienced research staff. Anthropometric measurements were taken at the start of the pre-test session, before the standardized warm-up. Standing height was measured using a fixed stadiometer (Holtain Limited, Crosswell, United Kingdom) to the nearest 0.1 cm, and body mass was measured to the nearest 0.1 kg using a digital scale (BC-554 Ironman Body Composition Monitor, Tanita, Illinois, USA). The standardized warm-up included 5 min of low-intensity running, 5 min of dynamic stretching, and progressive accelerations (four submaximal sprints, progressing to 90% of the players’ self-perceived maximal speed over a 30 m distance). Participants were also encouraged by supervisors to exert maximal effort during the physical tests.

Sprint test

The sprint test was conducted according to established guidelines19. Participants completed three trials each for 10 m, 20 m, and 30 m sprint, with a 20-second recovery between sprints and a 3-minute break between trials. All sprints began from a standing start, with the lead foot positioned 30 cm behind the start line and initiated after a 3-second countdown, to standardize the acceleration phase. Sprint time was recorded using an electronic twin beam photocell system (Witty, Microgate, Bolzano, Italy), and the fastest time was selected for analysis. The intraclass correlation coefficient (ICC) for the sprint times across the three trials in this study was 0.92–0.96. Previous studies have also demonstrated high reliability for the Witty photocell timing system, with ICCs ranging from 0.88 to 0.98 in short-distance sprint measurements20. The photocell timing gates were positioned 0.8 m above the ground, aligned approximately with the participants’ hip level. This placement minimized early triggering by limb or arm movement during acceleration and ensured consistent signal detection across trials, as recommended in previous reliability studies21.

Vertical jumping tests

The Countermovement Jump (CMJ) was used to evaluate jump height in centimeters22. Flight time from takeoff to landing was recorded using an optical timing system (Optojump, Microgate, Bolzano, Italy). Participants were instructed to perform maximal jumps. During the CMJ assessment, participants kept their hands on their hips and performed vertical jumps while maintaining this position. Takeoff actions were closely monitored, and unnecessary preparatory movements were not permitted. Individuals failing to adhere to these criteria were asked to repeat the test. Each participant performed three jumps, with 1-minute intervals of passive rest between jumps. The highest jump was record as the final score. Reliability analysis of CMJ data yielded an ICC of 0.95 in our sample. The Optojump system used in this study has been validated in previous research, showing high test-retest reliability (ICC > 0.90) for jump height and flight time measurements23.

Repeat-sprint ability test

The RSA test consisted of six 40-meter sprints, each separated by 20 s of passive recovery24. This design was selected because it requires acceleration, deceleration, a 180° COD, and re-acceleration, thereby reflecting basketball-specific movement demands. Although the distance slightly exceeds the 28 m court length, this format has been widely used in team-sport RSA research, including recent systematic reviews25,26.The test employed two pairs of infrared photocell timing gates (Witty, Microgate, Bolzano, Italy) positioned at the starting line and 20 m away. Participants began each sprint 30 cm behind the starting line, ran at maximum effort through the gates to the 20-meter mark, turned quickly, and sprinted back. Timing stopped when participants crossed the starting line. Athletes stood 30 cm behind the starting line before each sprint, which started after a 3-second countdown. This cycle was repeated six times with 20-second intervals between sprints. Sprint times were recorded, and two metrics were calculated: average time (RSAavg) and best time (RSAbest), both measured in seconds to two decimal places. The fastest time among the six sprints was documented as the peak performance. To evaluate fatigue, the Fatigue Index (FI) was calculated by normalizing the difference between the best and worst times against the best time, then multiplying by 100. The formula is as follows:

$$FI = 100 \times \frac{{{\text{worst}}\:{\text{time}}\left( s \right) - {\text{best}}\:{\text{time}}\left( s \right)}}{{{\text{best}}\:{\text{time}}\left( s \right)}}$$

Maximal oxygen uptake test

The Maximal Oxygen Uptake Test was conducted using a modified Bruce protocol, whose reliability and validity have been confirmed in previous research27. The protocol consists of seven 3-minute stages, with incremental increases in treadmill speed and incline at each stage. It begins at 2.7 km/h with a 10% incline, with both speed and incline increasing in subsequent stages. Throughout the incremental protocol, gas exchange and cardiac function were continuously monitored in real-time using a metabolic analyzer (Cortex Biophysik GmbH, Leipzig, Germany). Participants wore a heart rate monitor (POLAR H10, Finland) during the test, which was performed on a treadmill (Runner RUN 7410, RUNNER, Italy). The test was considered maximal and terminated when participants met at least three criteria from physiological and perceptual domains: (1) oxygen consumption plateau or decline across consecutive stages despite increased workload, (2) respiratory exchange ratio (RER) ≥ 1.15, (3) heart rate ≥ 180 beats·min−1 or failing to increase further within 2 min, and (4) Borg RPE approaching 19 accompanied by volitional exhaustion despite strong verbal encouragement. This combined approach ensured that objective physiological markers were prioritized, while subjective exertion served as an additional safeguard of maximal effort.

zig–zag COD speed test

The zig–zag test consists of four 5-meter sections (totaling 20 m) with cones placed at a 100° angle. The test was conducted using two pairs of photocell timing gates (Witty, Microgate, Bolzano, Italy). Participants were required to rapidly decelerate and accelerate while navigating around each cone. The test began from a stationary position, with the lead foot positioned 30 cm behind the first pair of photocell timing gates. Athletes received instructions before the test and must complete it expeditiously by crossing the second pair of timing gates. Each participant was allowed up to three trials, with a 3-minute rest period between each attempt. Of the zig–zag COD trials, the fastest completion time was used for analysis, consistent with standard COD performance testing protocols28. The ICC for zig–zag test performance in our sample was 0.91. This is consistent with previous findings demonstrating the zig–zag test’s strong test-retest reliability and construct validity in basketball and other team-sport athletes29,30. Furthermore, a recent systematic review confirmed the overall reliability and validity of agility assessments in team-sport contexts, supporting the methodological soundness of using the zig–zag test in the present study31.

Training program

During the pre-season period, an intervention program was implemented with a 6-week training protocol consisting of 12 sessions. Two training sessions per week, each lasting approximately 60 min, included warm-up, intervention training, and recovery. Before each session, participants completed a standardized 20-minute warm-up routine before each session, consisting of jogging, dynamic stretching, progressive running, COD training, and low-intensity explosive jumping. A minimum 48-hour recovery period was provided between sessions. During the first five weeks, week-to-week progression was guided by the principle of progressive overload, with gradual increases in sprint repetitions and total distance to stimulate neuromuscular and metabolic adaptation while preserving exercise quality. Beginning with a moderate initial load and progressing incrementally was also intended to support adherence and reduce the risk of dropout that can occur if initial intensity is set too high. In the final week, training distance was gradually reduced. This approach followed tapering principles, involving a progressive reduction in training load to decrease accumulated fatigue while maintaining training adaptations. Although the taper in this study was not individualized or periodized according to classical model (e.g., exponential or step taper), it was implemented as a brief, volume-based taper consistent with recommendations in high-intensity team-sport literature32. The primary aim was to minimize residual fatigue during the post-testing phase rather than peak for competition. Both groups maintained consistent distance and effort per repetition, session, and week. Participants were instructed to exert maximal effort during each session, which included 20-meter sprints. The main difference between the groups was the sprint trajectory: the EXP group performed 10 m + 10 m COD sprint with a sharp 180° turn, whereas the CON group performed linear sprints. After each sprint, both groups had 20 s of passive recovery and 4 min of rest between sets. Each warm-up and training session was supervised by a certified strength and conditioning coach, who provided verbal feedback and motivated players to give maximal effort. Training details were shown in Table 1.

Table 1 descriptive characteristics of the sprint training program performed by both experimental groups (RSA-LINEAR vs. RSA-COD).
Full size table

Statistical analyses

Data were analyzed using SPSS 26.0 statistical software (Chicago, IL, USA, version 26.0). Descriptive statistics were reported as mean ± standard deviation (Mean ± SD). The normality of the data was assessed using the Shapiro-Wilk test. Data were analyzed using a 2 (time: pre vs. post) × 2 (group: EXP vs. CON) mixed-design ANOVA. When significant interaction or main effects were detected, Bonferroni-adjusted pairwise comparisons were conducted to examine simple main effects. This correction was applied consistently across all post-hoc analyses to control for Type I error. In this analysis, the partial eta squared (ηp2) was calculated and interpreted as follows: ≥ 0.01 indicates a small effect, ≥ 0.059 a medium effect, and ≥ 0.138 a large effect. Cohen’s d was computed to assess differences between and within groups, with effect sizes interpreted as < 0.2 for trivial, 0.2–0.6 for small, 0.6–1.2 for medium, and > 2.0 for large effects33. Statistical significance was set at p < 0.05.

Results

Table 2 displays the descriptive statistics, Mixed-design ANOVA outcomes, effect sizes, and alterations in individual data and mean values pre- and post-intervention for all assessments. No notable variances were observed in the baseline data across all tests.

Table 2 Performance outcomes at baseline and after the intervention period for both groups.
Full size table

Sprint performance

No significant time × group interaction effects were found for the 10 m (F = 0.387, p = 0.54, ηp2 = 0.014, medium effect), 20 m (F = 0.116, p = 0.74, ηp2 = 0.004, small effect), or 30 m (F < 0.001, p = 0.99, ηp2 = 0.000, small effect) sprint tests, indicating that the changes over time did not differ significantly between the two groups. Similarly, no main effects of group were detected for any sprint distance (p > 0.05). However, a significant main effect of time was observed for the 10 m sprint, indicating an overall pre–post improvement (F = 7.025, p = 0.01, ηp2 = 0.201, large effect). In contrast, the time × group interaction was not significant, and the within-group post-hoc comparisons did not reach significance (EXP: p = 0.09; CON: p = 0.31). Accordingly, there was no evidence to support a group-specific intervention effect. For the 20 m (F = 1.650, p = 0.21, ηp2 = 0.056, small effect) and 30 m (F = 0.033, p = 0.85, ηp2 = 0.001, small effect) sprint tests, no significant main effects of time were detected.

Vertical jumping performance

No significant time × group interaction was found for CMJ height (F = 0.065, p = 0.80, ηp2 = 0.002, small effect), and no main effect of group was detected (F = 0.517, p = 0.48, ηp2 = 0.018, small effect). However, a significant main effect of time was observed (F = 4.495, p = 0.04, ηp2 = 0.138, medium effect), indicating an overall improvement in CMJ performance when collapsed across groups. In contrast, within-group post-hoc comparisons did not reach statistical significance (EXP: p = 0.77, d = 0.72; CON: p = 0.53, d = 0.47). Accordingly, these findings support a time-related improvement in vertical jump performance, with no evidence of a differential effect between the two training protocols.

Repeated sprint performance

Significant time × group interaction effects were observed for RSA average time (RSAave; F = 6.954, p = 0.01, ηp2 = 0.199, large effect) and fatigue index (RSAFI; F = 4.369, p = 0.05, ηp2 = 0.135, medium effect). No main effects of group were found (p > 0.05), but significant main effects of time were detected (RSAave: F = 4.282, p = 0.05, ηp2 = 0.133, medium effect; RSAFI: F = 9.888, p < 0.01, ηp2 = 0.261, large effect). Post-hoc analyses indicated that improvements in RSAave and RSAFI were greater in the EXP group (p = 0.01, d = − 1.08; p = 0.02, d = − 1.15) compared to the CON group, which also improved but to a smaller extent (RSAave: p = 0.44, d = 0.15; RSAFI: p = 0.92, d = − 0.22). In contrast, no significant interaction effect was found for RSAbest (F = 0.064, p = 0.80, ηp2 = 0.002, small effect), nor were there significant main effects of group (F = 0.484, p = 0.49, ηp2 = 0.017, small effect) or time (F = 2.441, p = 0.13, ηp2 = 0.080, medium effect). These findings suggest that COD-based repeated sprint training confers greater benefits in repeated sprint ability and fatigue resistance, but does not provide additional improvements in best sprint performance.

Maximal oxygen uptake performance

A significant time × group interaction effect was observed for VO2max (F = 5.917, p = 0.02, ηp2 = 0.174, large effect), along with significant main effects of group (F = 4.439, p = 0.04, ηp2 = 0.137, medium effect) and time (F = 9.886, p < 0.01, ηp2 = 0.263, large effect). Bonferroni-adjusted comparisons revealed no significant difference between the EXP and CON groups at post-test (p = 0.61). However, post-hoc analyses showed divergent within-group improvements: the EXP group demonstrated a significant increase (p = 0.01, d = 1.15), whereas the CON group exhibited only a negligible change (p = 0.98, d = 0.22). These results indicate that while post-test VO2max levels did not differ significantly between groups, the magnitude of improvement over time was greater in the EXP group.

Zigzag COD speed performance

A significant time × group interaction was observed for zig–zag performance (F = 5.201, p = 0.03, ηp2 = 0.157, large effect). No significant main effect of group was detected (F = 0.436, p = 0.51, ηp2 = 0.015, small effect), while a significant main effect of time was found (F = 7.102, p = 0.01, ηp2 = 0.202, large effect). Post-hoc analyses indicated that the EXP group demonstrated significantly greater improvements in COD speed compared with the CON group. Effect size analysis further supported this finding, showing a large improvement in the EXP group (p = 0.02, d = − 1.07) but negligible change in the CON group (p = 0.34, d = − 0.20).

Discussion

The present study aimed to evaluate the effects of RST with COD on physical performance in college female basketball players. The findings indicate that RST with COD resulted in greater improvements in RSA, VO2max, and agility (zigzag test performance) compared with linear sprint training, as evidenced by significant interaction effects. Improvements in short-distance sprint performance were observed in both groups; however, no differential effect could be attributable to the intervention. Although improvements in some performance indices (e.g., CMJ, RSAave, RSAFI) were observed in both the experimental and control groups, the most pronounced differences between the groups emerged for RSA, VO2max, and COD measures. These outcomes warrant careful discussion in the context of previous research and highlighting potential limitations and directions for future investigation.

Both the EXP and CON groups showed decreased 10-meter sprint times after the intervention period. This improvement was statistically significant over time (p = 0.01), indicating a general training effect. The lack of a significant time × group interaction (p = 0.54) suggests that neither training protocol was superior in enhancing short-distance sprint performance. These results suggest that both RST with COD and traditional linear repeated sprint training can improve acceleration capacity, possibly through increased neuromuscular efficiency. In addition, learning effects from repeated test exposure, such as greater familiarity with the start procedure, movement preparation, and acceleration pacing, may also have contributed to the observed improvements. To minimize this bias, two familiarization sessions were conducted before the formal pre-test, standardized warm-ups and instructions were provided, and the fastest of three trials was retained for analysis, under identical conditions for both groups. Thus, any learning effects were likely comparable across groups. Consistent with previous research, conditioning based on repeated sprints seems to benefit short-distance speed, regardless of the trajectory type16,34.

Incorporation of COD elements during RSA yielded notable benefits for agility, as evidenced by the significant enhancement in zig–zag test performance in the EXP group. This finding aligns with prior research highlighting the importance of rapid directional changes in basketball performance, which can be enhanced through targeted drills35. Arslan et al.36 reported that training programs incorporating high-intensity interval elements as opposed to small-sided games can exert distinct effects on agility and COD speed36. The observed improvements can be partially attributed to the principle of training specificity, which suggests that performance enhancements are maximized when training closely mimics the movement patterns, velocity profiles, and muscular requirements of the target task37. However, recent work highlights that neuromuscular adaptation involves not only task-specific similarity but also broader mechanisms such as neural drive, coordination, and stretch–shortening cycle efficiency. Cherni et al.11 further demonstrated that even COD-based interventions can elicit distinct adaptation patterns depending on the mechanical load (loaded vs. unloaded plyometric training), underscoring the complexity of neuromuscular responses. Our study showed that the zigzag test required multidirectional accelerations, decelerations, and rapid directional changes - all of which were directly addressed in the COD-based repeated sprint regimen. This significant biomechanical and neuromuscular similarity likely contributed to the performance enhancements observed, underscoring the importance of replicating sport-specific movement patterns during training. Beyond task specificity, preparatory routines may also exert acute biomechanical effects on subsequent performance. Martone et al.38 reported that dynamic stretching and core stability exercises can enhance neuromuscular readiness by improving joint range of motion, trunk stiffness, and proximal stability, thereby optimizing distal force transmission during accelerations and braking actions38. Consistent with this perspective, other studies have demonstrated that specific dynamic warm-up protocols, including those incorporating weighted vests, can acutely enhance repeated change-of-direction performance in team-sport athletes3940. Since the warm-up protocols were standardized and identical across groups in our study, these acute effects are unlikely to confound between-group comparisons; nevertheless, they provide valuable context for understanding how preparatory strategies may have interacted with specificity principles to facilitate COD-related improvements. Furthermore, sex-specific adaptations warrant explicit attention. Although the present study included only female collegiate basketball players, the methodological framework outlined by Elena Mainer Pardos et al.4040, which was developed in male athletes, provides useful insights into potential adaptation mechanisms associated with sprint-based training. However, because these findings were derived from male populations, their direct applicability to female athletes should be interpreted with caution. Their findings suggest that moderators such as tendon stiffness, neuromuscular recruitment strategies, and eccentric load tolerance may differ between sexes, thereby influencing the adaptation trajectories to COD versus linear sprint training40. Incorporating this perspective situates our results within a broader context and underscores the importance of future research aimed at examining sex-specific responses to sprint-based interventions.

RSA performance is a crucial determinant of success in team sports. Our study showed that the EXP group demonstrated significant short-term improvements in average RSA time and fatigue index. These findings suggest that incorporating COD components improves immediate neuromuscular responses and enhances fatigue resistance during repeated sprint sequences. Previous research in soccer has similarly shown that repeated sprint training enhances RSA capacity41. Moreover, Spencer et al.42 reported a strong correlation between fundamental fitness attributes, including RSA, and overall performance in young football players42.

Interestingly, RSAbest did not exhibit a statistically significant improvement in either group, despite clear gains in RSAave and RSAFI. A possible explanation is that the training stimulus primarily targeted fatigue resistance and sustained sprint performance rather than maximal single-effort output. Additionally, although participants were verbally encouraged to perform each sprint maximally during training, no objective measures (e.g., peak velocity via GPS or power monitoring) were employed to verify maximal effort. This limitation raises the possibility that some athletes may not have consistently reached true maximum sprint intensity during training, potentially explaining the absence of RSAbest improvement. Future research should consider integrating real-time intensity monitoring tools to ensure maximal output during high-intensity sprint training and better distinguish adaptations in fatigue resistance versus maximal neuromuscular performance. Moreover, recent evidence indicates that improvements in RSA are closely linked to changes in the reactive strength index and leg stiffness. Włodarczyk et al.43 demonstrated that sprinters who improved their RSA performance often exhibited concurrent enhancements in RSI and optimized stiffness regulation, both of which facilitate more efficient force transmission and faster transitions from eccentric braking to concentric propulsion43. In the context of our study, COD-RST, which involves frequent decelerations followed by explosive re-accelerations, likely provided a stronger stimulus for developing these qualities compared with linear sprinting. This mechanistic pathway may therefore offer a reasonable explanation for the differential adaptations observed in RSA indices between the two training modalities.

The EXP group demonstrated significant improvements in aerobic performance, as assessed by VO2max, indicating enhanced cardiovascular and metabolic efficiency. These results align with previous research in high-intensity team sports training, such as Iaia et al.44, which showing that high-intensity training protocols lead to notable improvements in aerobic fitness44. Additionally, Buchheit et al.45 observed that supramaximal training protocols can result in substantial increases in aerobic capacity45. Although such improvements are often attributed to combined central cardiovascular (e.g., stroke volume, cardiac output) and peripheral muscular (e.g., oxidative capacity) adaptations, our study did not directly measure these parameters, and therefore the precise mechanisms cannot be confirmed. Nevertheless, the findings suggest that COD-based repeated sprint training may enhance aerobic capacity in female basketball players, potentially facilitating faster recovery between high-intensity efforts and reducing fatigue during competition. Future studies should incorporate direct physiological assessments to further clarify the mechanisms underlying VO2max improvements.

Improvements in CMJ performance were observed in both training groups, despite the lack of a significant time × group interaction. This outcome likely reflects shared neuromuscular adaptations between repeated sprint training and vertical jumping. Sprinting and jumping both rely heavily on lower-limb explosive strength, rapid force production, and efficient utilization of the stretch-shortening cycle (SSC)46. Sprint efforts, particularly those involving maximal acceleration phases, require high activation of fast-twitch muscle fibers, which are also crucial in vertical jump execution. Furthermore, repeated sprinting under load may enhance neuromuscular coordination, increase neural drive to the working muscles, and improve lower-limb stiffness—all of which contribute to more efficient energy transfer during SSC-based movements. Previous studies have noted similar crossover effects between sprint and jump training, even when jumping was not specifically targeted47,48. These shared physiological pathways likely explain why both the COD-based and linear sprint training protocols were effective in enhancing vertical jump performance. Additionally, repeated near-maximal sprinting may have induced post-activation performance enhancement (PAPE). As highlighted by Formiglio et al. (2024)50, the balance between potentiation and fatigue is determined by factors such as exercise intensity, recovery duration, and athlete characteristics. In our training design, repeated short sprints with standardized work-to-rest ratios likely provided sufficient neural stimulation under conditions favoring potentiation, thereby accelerating the rate of force development and further contributing to the CMJ improvements observed in both groups.

This study has several limitations. The sample size was relatively small and restricted to female collegiate basketball players, which limits the generalizability of the findings and may introduce selection bias. Regarding the outcomes, although the intervention produced significant improvements in some measures, sprint distances and RSAbest did not show notable changes, suggesting that the training protocol may require further refinement. Future research should employ larger sample sizes and multi-center designs to better explore dose–response relationships. Moreover, training intensity was not objectively monitored during sessions, which limits our ability to verify that the intended maximal efforts were consistently achieved. To improve training-load control, future studies should integrate real-time monitoring tools (e.g., GPS, timing gates) during training sessions to quantify session intensity and execution quality and better standardize training load. Regarding external factors, this study did not sufficiently control for nutritional status and fatigue management. Given the importance of energy availability for female athletes’ performance, future studies should more rigorously monitor dietary intake and recovery strategies. In terms of testing methodology, potential learning or test adaptation effects could not be fully excluded. This consideration is particularly relevant when interpreting the changes observed in the 10 m sprint performance. Although the control group showed a slight numerical decrease in 10 m sprint time, this change did not reach statistical significance. Therefore, the significant main effect of time observed in the analysis was primarily driven by the improvements in the experimental group rather than a consistent improvement across both groups. Furthermore, the study design lacked a true control group that participated only in routine basketball training without any supplementary RST. The inclusion of such a group in future research would help isolate the specific effects of the RST interventions from general training adaptations. Finally, although no injuries occurred during the intervention, the high-intensity nature of COD training—with its repeated decelerations and rapid directional changes—carries an inherent risk of injury. From a broader biomechanical perspective, recent frameworks emphasize the importance of integrating performance enhancement with injury risk mitigation in sport training contexts49. This study did not systematically evaluate injury monitoring or prevention strategies, and this omission limits the practical applicability of the findings for training implementation.

Conclusion

This study demonstrates that RST-COD produces greater improvements in RSA, VO2max, and COD speed than linear sprint training in female collegiate basketball players. Although both training methods improved short-distance sprint performance, no significant interaction effect was found, suggesting that these improvements cannot be attributed solely to the COD-based protocol. These findings highlight the effectiveness of COD-specific repeated sprint drills in enhancing agility and fatigue resistance, both of which are critical components of basketball-specific fitness. Coaches are encouraged to incorporate such protocols into conditioning programs to better prepare athletes for the multidirectional and intermittent demands of competitive basketball.